1. October 2011 David Toback, Texas A&M University Research Topics Seminar1 David Toback Texas A&M University University of Chicago Seminar November 2015

2. Prologue UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF2

3. Outline UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF3

4. Top Production at the Tevatron UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF4

5. Definition of A FB UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF5 In proton-antiproton collisions can measure the relative angle between the outgoing top and the proton direction This is a forward event… If the top was going in the direction of the anti-proton  backward event Forward-Backward Asymmetry: A FB = Fraction(Forward) – Fraction(Backward)

6. Quick Aside on Variable Choice  y June 2012 David Toback, Texas A&M University CIPANP, A fb in Top Quarks at CDF6 Note: y doesn’t have the usual geometric angle many of us are used to. At hadron colliders we usually use pseudo-rapidity which assumes m=0  Here E and p not close to equal because of the top mass Transform from  t to rapidity (y) Rapidity difference is a good proxy for production angle Invariant under longitudinal boosts Asymmetry in  y is bigger than in  t, easier to distinguish from 0

7. Tevatron VS. LHC Isn’t the LHC better for this? Tevatron is proton-antiproton while LHC is proton-proton Tevatron provides a direction in space, while LHC only provides a line in space Can directly use the opening angle relative to learn about Feynman diagrams –Can get at the same diagrams at LHC, but takes much larger statistics and more indirect methods UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF7

8. Asymmetry Predictions First SM predictions of A FB after top discovery (~1995) No A FB at LO, all asymmetry comes from NLO diagrams. Took many years to get decent predictions –Interference among diagrams –Larger-than-expected EW correction and higher order QCD corrections complicate the calculation Note: Process has been a harbinger of things to come: We have learned during the process that LO MC’s are simply not good enough to measure properties of the top quark UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF8

9. Lots of Predictions UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF9

10. Comments on Systematics & Biases The measurements presented today are statistics limited, so focus on techniques which make the most of the data Spent a lot of time worrying that our methods aren’t biased Won’t rehash the results from Lep+Jets analysis as those were published a few years ago… UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF10

11. Selecting Top Quark Events UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF11 Lepton+Jets Final State Dilepton Final State 1 reconstructed lepton Missing transverse energy ≥4 jets (1 b-tag)  E T > 220 GeV Bigger branching fraction, more final state particles to measure, bigger backgrounds 2 reconstructed leptons Missing transverse energy ≥2 jets  E T > 200 GeV Higher purity sample, but smaller branching fraction, two leptons have better angles, but two neutrinos cause reconstruction ambiguities

12. Inclusive A FB Results UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF12 Prediction in plot is old so one can see where the excitement came from Can already tell that the asymmetry is larger than expected for large |  y|

13. CDF Dilepton Result with 5.1fb -1 UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF13

14. Moving Forward Adopt a quasi-historical approach from here on out For awhile there was a the real sense that there was a significant discrepancy between the data and the SM predictions Goals: –Improve SM prediction –Consider other ways to look at the data –New physics models? UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF14

15. Recurring Theme: Lots of Ways to Measure and/or Compare to predictions UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF15

16. A FB as a function of  y Observe larger asymmetry at larger values of  y Differential values of A FB well described by a linear relationship –Not clear to theorists why this should be so… Slope: 0.253±0.062 –PRD 87, 092002 (2012) 2.2 higher than NNLO SM prediction –Slope: 0.114 +0.005 -0.012 –PRL 115, 052001 (2015) June 2012 David Toback, Texas A&M University CIPANP, A fb in Top Quarks at CDF16

17. UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF17

18. UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF18 Quantify with a Slope: (15.5±4.8)x10 -4 Observed (3.4±1.2)x10 -4 NLO Not a huge anomaly, but certainly we notice it and push forward… Can the numerical values of these slopes be useful to model builders?

19. UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF19 Examples of New Physics that could give a Large A FB Two main classes of models S-channel mediator –e.g. axigluon T-channel flavor changing mediator –e.g. W’ or Z’ Although many of these have strong constraints (and many are now excluded), they provide a good model for searches For a review see: M. Gresham, I.-W. Kim and K. Zurek, Phys. Rev. D83 114027 (2011 )

20. Fun historical aside… This is when my team joined the hunt Before we joined: UChicago tie-in Amidei and his group led Lep+Jets –His postdoc Jon Wilson now works with me Top  Dileptons datasets were a mess Ziqing Hong, student who worked with me, cleaned up dilepton datasets, then systematically did all the measurements Start picking them off one-by-one… UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF20

21. Leptonic A FB Definition UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF21 SM: A FB = 3.8%, specific ratio for Reconstructed/Leptonic Various BSM Models: -5% < A FB < 15%, ratio can vary significantly

22. Dilepton-only Observable UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF22 Same BSM models: -9% < A FB < 22%

23. Quick Aside on the Method The methodology for measuring the Leptonic asymmetry turned out to be both simple and complicated Not sure why it worked Spent some real time trying to understand why Turned out to be so simple that I thought I’d share what we learned UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF23

24. Leptonic Methodology UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF24

25. Clever idea: Break into parts UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF25

26. Why does it work? UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF26

27. Fun details UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF27 If we look at the contribution to the asymmetry as a function of , we see that for different means (total A FB ) we get very different shapes However, the shapes basically look the same, just with a different overall normalization –Gives us a robust way to measure things Interesting: –Very little sensitivity at the center of the detector/where most of the events are –Most of the sensitivity comes around |  |=1

28. More on how well it worked UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF28 Works well as a function of q  Works well for many different input values of A FB

29. Leptonic A FB from Lepton+Jets UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF29

30. Leptonic A FB from Dileptons UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF30 Measure A FB with CDF full dataset in dilepton (9.1 fb -1 ) A FB = 0.072±0.060 –A FB (SM,NLO) = 0.038 ±0.003 Combine CDF measurements  Result is 2  larger than NLO SM prediction: –A FB = 0.090 ± 0.028 Published as PRL 113, 042001 (2014) (CDF)

31. Measurement techniques works equally well for A FB in  –A FB =0.076±0.072(st)±0.039(sys) = 0.076± 0.081 Result consistent with SM –A FB (SM,NLO) = 0.048 ± 0.004 –PRL 113, 042001 (2014) UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF31 Leptonic  Results (Dileptons only)

32. State of the Art for Leptonic A FB All the results, including from DZero, are consistent with SM predictions Notice that all of them are above… UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF32

33. Transition to Reconstructed A FB Taking Stock of where we are, and looking forward Observed large A FB in reconstructed Lep+Jets, but only big in differential measurements Looked at Leptonic A FB : Still above predictions, but not anomalously so Next: Even though we won’t have enough statistics to say anything conclusive we measure the full reconstructed A FB in dileptons UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF33

34. Reconstruction in Dileptons UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF34

35. How well does the Reconstruction Do? UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF35

36. Resolution Continued Consider all solutions and weight them according to their probability from the fit Find 61% have a y t measured within 0.5 of truth value UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF36

37. Going from Reconstruction to a Measurement of A FB Need a sophisticated methodology to turn reconstructed  y measurement into a measurement of  y at the parton level Want to measure the inclusive A FB as well as the differential A FB vs.  y. Correct for two effects: 1)Not able to measure all events Limited detector coverage Imperfect event selection efficiency 2)Not able to measure y t correctly for events we do have a measurement Finite detector response resolution Imperfect tt momenta reconstruction UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF37

38. Optimization Slide 1 UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF38 Tried to do something state-of-the- art since this is a legacy measurement Optimize before looking at data Minimize the expected uncertainty on A FB 1) Use more information in the measurement –Keep full probability distribution rather than pick the most probable solution –Weight both lep-jet pairings with likelihoods –Add information from b-jet charge

39. Optimization Slide 2 UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF39 2) Reject tt with low reconstruction quality –Jet energy dragged too far from measured values to make a tt –m 2 lb too high, not likely a good top –Lepton lying on top of a jet, most likely a W+Jets Expected uncertainty goes from 0.144 before optimization and 0.122 after

40. Optimization Results UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF40

41. Validation/Unfolding Results To extract parton- level A FB, run MCMC to find most probable parameters that match observation Extract A FB from marginalized posterior distribution POWHEG sample with 10M events gives 0.053±0.013 with 0.0524 generated UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF41

42. Comparing MC to Truth UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF42

43. Reconstruct/Unfolding Results Results with dilepton data: –A FB = 0.12±0.11(stat) ±0.07(syst) = 0.12±0.13 Combined with CDF result in lepton+jets –A FB (CDF) = 0.160±0.045 Consistent with prediction –A FB (N 3 LO SM) = 0.100±0.006 within 1.5  Manuscript in preparation, to be submitted to PRD UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF43

44. A FB Reconstructed Results UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF44

45. Tevatron Results Final individual results on A FB from Tevatron All results consistent but higher than NLO (and NNLO) SM predictions A combination of CDF+DZero results is in the works UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF45

46. Summary and Conclusions UC Seminar, Nov 2015 David Toback, Texas A&M University A FB in Top Quarks at CDF46

47. Backups June 2012 David Toback, Texas A&M University CIPANP, A fb in Top Quarks at CDF47